AI Product Engineer building practical AI applications, evaluation workflows, and backend systems that turn vague goals into measurable progress.
Toronto · Taipei
Personal site · Writing · GitHub
I build software for the gap between "I know what I should do" and "I actually did it."
That gap shows up everywhere I build: system design interview practice, French speaking preparation, personal execution, and engineering teams trying to use AI without turning their workflow into fog.
My instinct is to turn messy progress into a system: practice loops, scoring, evidence, history, privacy boundaries, recovery paths, dashboards, and enough observability for the product to explain itself.
My background is backend engineering, so I do not think of AI products as chat boxes. I think of them as workflows with state, evaluation, failure modes, cost limits, user trust, and measurable improvement.
These days I am focused on one question:
How do we build AI products that help people improve at hard things, instead of just generating more text?
| Product | What it demonstrates |
|---|---|
| SpeakUp SD | AI system design interview practice with speech evaluation, whiteboard evidence, scoring reports, readiness checks, deployment operations, and customer trust surfaces. |
| TCF Canada | AI speaking coach for French immigration exam preparation with speaking-practice UX, product positioning, language-learning flows, and a deployed customer-facing experience. |
| Project | What it demonstrates |
|---|---|
| 12wy-tracker | Local-first Tauri desktop app for 12 Week Year execution, BYOK/Ollama AI coaching, prompt logs, budget limits, and eval artifacts. |
| commit-time-machine | Browser-based commit pattern analyzer for cadence, timing, streaks, and readable engineering telemetry. |
| ai-workflow | Practical AI-agent workflow templates, prompts, and operating notes from a senior backend engineering perspective. |
Before building AI applications, I worked on production backend systems: real-time communication, IoT device platforms, multi-tenant workflows, APIs, observability, and infrastructure reliability.
That experience shapes how I build AI products: not as shiny demos, but as systems with feedback loops, failure paths, privacy boundaries, cost awareness, and measurable user progress.
- Building AI workflows that fit real engineering teams and production constraints.
- Studying system design, evaluation, language-learning products, and AI-assisted engineering practice.
- Writing at Zoe Site about backend systems, AI product engineering, and plain-language technical decisions.
Contributions 1327 total · 71 public commits · 1220 private
First activity 2019-04-28
Active span 7.2 years · 85/2616 days · 3.2%
Cadence favorite day SAT · weekend 41% · avg 15.6/active day
Streaks current 1 days · best 17 days
Best year 2026 (1020)
Signals streak starter · century · 500 club · kilo contributor · weekend builder · library owner · polyglot · veteran
Auto-updated daily. Private contributions are counted by GitHub summary only; peak timing uses public commits.